-2σ

+-2sigma.Rmdのロジックを良くしたい
やりたいこと
mm/ddのリターンが過去〇〇日のリターンの-2σ以上なのか以下なのか判別したい

1 前処理

2 -2σ計算

rm(tmp)
rm(tmp1)
rm(tmp2)
rm(touraku2)


#過去1週間分を評価する
  #結果テーブル準備
    # tmp2 <- data.frame('date'=2017-01-01 , 'close'=999 , 'ticker'='aaa' , 'rtn'=0.99 ,
    #                        'sigma'=0.99 , 'buysign'='buy')[-1,]
    # tmp2 <- as.Date(tmp2[,1])
    # touraku2 <- data.frame('date'=2017-01-01 , 'close'=999 , 'ticker'='aaa' , 'rtn'=0.99 ,
    #                        'sigma'=0.99 , 'buysign'='buy')[-1,]
    # touraku2[,1] <- as.Date(touraku2[,1])
    tmp2 <- dailyrtn
    tmp2 <- transform(tmp2, sigma=0)
    tmp2 <- transform(tmp2, buysign='--')
    tmp2[,1] <- ymd(tmp2[,1])
    tmp2[,6] <- as.character(tmp2[,6])
    tmp2 <- tmp2[-1,]
    touraku2 <- head(dailyrtn,1)
    touraku2 <- transform(touraku2, sigma=0)
    touraku2 <- transform(touraku2, buysign='--')
    touraku2[,1] <- ymd(touraku2[,1])
    touraku2[,6] <- as.character(touraku2[,6])
    touraku2 <- touraku2[-1,]

  for (i in 1:length(ticker_tmp)){
    
    tmp <- subset(dailyrtn, ticker == ticker_tmp[i])
    #列追加
      tmp <- transform(tmp, sigma=0)
      tmp <- transform(tmp, buysign='--')
      tmp$date <- ymd(tmp$date)
      tmp[,6] <- as.character(tmp[,6])
      tmp <- tmp[-1,]

    #1週間前の位置 1週間前から現在までのcloseを評価するの意
      tarRow1 <- nrow(tmp) - 6
      
      for (k in tarRow1:nrow(tmp)) {
      #過去300日分のリターンの-2σ計算
        #tarRow2 <- k - 300
        #過去300日分の抽出
          #tmp1 <- subset(tmp, 
          #               as.numeric(row.names(tmp)) >= tarRow2
          #               & row.names(tmp) <= tarRow1 )
          tmp1 <- tmp[1:tarRow1,]
          tmp1 <- tail(tmp1, 300)
        #-2σ計算
          under2sigma <- quantile(tmp1$rtn,c(0.02275))
        #-2σの追記
          tmp[k, 5] <- under2sigma
        #リターンとunder2sigmaを比較してbuyサインを立てる
          if (tmp[k, 4] < tmp[k, 5]) {
            tmp[k, 6] <- 'buy'
          } else {
            tmp[k, 6] <- '-'
          }
      }
    #結果をテーブルへ
      tmp2 <- rbind(tmp2, tmp)
      touraku2 <- subset(tmp2, buysign=='buy')
      touraku2$ticker <- as.factor(touraku2$ticker)
      #1週間以内でフィルタ
        touraku2 <- touraku2 %>%
          filter(date >= Sys.Date()-7)
  }
  datatable(touraku2,
          filter = 'top', 
          style = 'bootstrap', class = 'table-bordered table-condensed',
          extensions = 'ColReorder',
          options = list(dom = 'Rlfrtip')
          )

  #write.table(touraku2,'touraku2.txt',sep="\t",quote=F,row.names = F)
  
  ticker_bakusagari <- as.character(t(touraku2$ticker))

#購入済で下がった銘柄は？
  ticker_kaimashikoho <- intersect(ticker_bakusagari, kounyu)
#購入済でない銘柄は？
  #setdiff(a,b)はaにあってbにないものを抽出する
  ticker_shinkikoho <- setdiff(ticker_bakusagari, kounyu)
  
  tmp <- subset(kabu.com_list, kabu.com_list$ticker %in% ticker_bakusagari)
  datatable(tmp,
        filter = 'top', 
        style = 'bootstrap', class = 'table-bordered table-condensed',
        extensions = 'ColReorder',
        options = list(dom = 'Rlfrtip')
        )

購入済で下がった銘柄
2510

新規購入候補の銘柄
1471, 1327, 2066, 3462

3 グラフ（確認用）

#国内

if (length(intersect(touraku2$ticker, kokunaiETF)) >= 1) {
  g <- ggplot(NULL)
  g <- g + geom_line(data = subset(daily1,
                                   ticker %in% intersect(touraku2$ticker, kokunaiETF) &
                                   year(date) >= 2017),
                     aes(x=date, y=close, colour=ticker))
  g <- g + geom_point(data = subset(touraku2,
                                    ticker%in% intersect(touraku2$ticker, kokunaiETF) &
                                    year(date) >= 2017),
                      aes(x=date, y=close, colour=ticker))
  ggplotly(g)
} else {
  tmp <- 0
}

#海外
if (length(intersect(touraku2$ticker, kaigai)) >= 1) {
  g <- ggplot(NULL)
  g <- g + geom_line(data = subset(daily1,
                                   ticker %in% intersect(touraku2$ticker, kaigai) &
                                   year(date) >= 2017),
                     aes(x=date, y=close, colour=ticker))
  g <- g + geom_point(data = subset(touraku2,
                                    ticker%in% intersect(touraku2$ticker, kaigai) &
                                    year(date) >= 2017),
                      aes(x=date, y=close, colour=ticker))
  ggplotly(g)
} else {
  tmp <- 0
}

4

  ggplotly(
      ggplot(subset(monthlyrtn, ticker %in% ticker_bakusagari),
             aes(x = return_monthly, colour = ticker)
             ) + geom_density()
  )

# 購入済銘柄との相関確認

ticker4 <- c()
f02 <- daily1 %>%
  filter(ticker %in% ticker_bakusagari | ticker %in% kounyu,
         date >= Sys.Date() - 365)
f02 <- dcast(f02, date ~ ticker, value.var='close')
f02 <- f02[,2:ncol(f02)]
f02 <- f02[,order(factor(colnames(f02),levels=ticker4))]
cor.plot(cor(f02, use='pairwise.complete.obs', method='p'),
         numbers=T)

corrplot.mixed(cor(f02, use='pairwise.complete.obs', method='p'),
               order = "hclust", tl.col = "black")

#qgraph
  f03 <- f02
  groups <- list("購入済"=1:length(kounyu),"新規"=length(kounyu)+1:length(ticker_shinkikoho))
  qgraph(cor(f03, use='pairwise.complete.obs', method='p'),groups=groups)

    $(function(){
      $("img:not(.lb-image)").wrap(function() {
        return "<a href='" + $(this).attr("src") + "' data-lightbox='" + $(this).attr("scr") + "'></a>";
      });
    });